An Improved Algorithm for Radar-derived Classification of Convective and Stratiform Precipitation Echoes

Scott W. Powell and Stacy R. BrodzikUniversity of Washington, Seattle, WA

MJO Field Data and Science Workshop, Kohala Coast, HIMarch 4, 2013

This work was supported by NSF grant AGS-1059611 and DOE grant DE-SC0008452.Breakout Session 1A

Why we need a new/improved algorithm

• Steiner et al (1995; SHY95) requires an interpolated dataset.

• A standard 2 km interpolation, or even higher resolutions, “creates” data. This is bad.

• Small, isolated convective echoes can be completely missed by SHY95.

Fixing the problem

• Use the native resolution of the dataset, taking advantage of the high resolution typical along radials.

• Easy gridding onto irregular Cartesian grid for determining distance between data points.

R (highest spatial resolution)

Φ1. (R,Φ) -> (X,Y)2. Determine distance between red box and all other data points. Yellow boxes represent those within specified distance.

Fixing the problem

• Use the native resolution of the dataset, taking advantage of the high resolution typical along radials.

• Easy gridding onto irregular Cartesian grid for determining distance between data points.

• Shallow convection scheme

Isolated Convective Echoes

Large Echo Clusters

Squall Line

Widespread Stratiform with Embedded Convection

TOGA

(Soon) Upcoming Improvements

• Beam blockage (for Addu radars)

- Mask out Addu City completely

- Eliminate second-trip echo.

- Update shallow convection scheme.

S-Pol

TOGA

Single Z-R (tentative)Z = 129.56 R1.37

Hybrid rain rate

Improved stratiformfraction during suppressed periods

Availability/Suggestions

• Only input currently accepted is CFRadial.

• Code currently only written for usage with MATLAB.

• How to make easy-to-use for most users?

• Other suggestions/concerns?